[Bioclusters] cluster hardware question

[Joseph Landman]

One of the "hidden" costs of clusters (or tin) is the management
aspect.  Few really people really grasp upfront that in a cluster, these
things are not automagically bundled, so you either need to pay someone
money on your staff, your suppliers staff, or call someone else.

Rarely if ever does anyone talk about cluster support up front.  It is
one of those hidden "indirect" costs.  It is the "cost of doing
business" in some cases.  

I find that a not-insignificant portion of my business is helping out
with the "how do I do this" sort of problem, or "the machine has some
blinkenlicht issues, can you help".  

Good cluster management is not (currently) trivial and good support
isn't cheap.  

These are good points you made.

Teaching people how to support is IMO much harder than teaching the
tin-pinning.  There is the aspect of the book learning "if the light
blinks than do this" and the reality of the highly upset users of the
machine not currently functional.  I am reminded of a joke about mission
focus in the a setting with alligators and swamps...

Building a production system means building a production support group. 
This is hard to do right.  Finding people who really know what can go
wrong, how badly, and hopefully how to fix it (or who to call to get the
knowledge).  The design part is not trivial.  The support part is hard.



On Fri, 2003-01-17 at 19:07, James Cuff wrote:
> On 16 Jan 2003, Joe Landman wrote:
> 
> > Pay Chris (Bioteam) and I (Scalable Informatics) to come out and have a
> > beer with you ... it will be a rather long beer....  :)
> 
> Pinning big bits of tin on to your network is only half the problem.
> It's the long term support that is the really hard bit.  However, lerning
> how to 'pin on the tin' is hard, and does need help.
> 
> J.
-- 
Joseph Landman, Ph.D.
Scalable Informatics LLC
email:   landman@scalableinformatics.com
  web:   http://scalableinformatics.com
voice:  +1 734 612 4615
  fax:  +1 734 398 5774